Poor powder flow leads to many problems during manufacturing and can lead to inaccurate dosing and offspecification products. Powder flowability is commonly assessed under relatively high applied loads using shear cells by characterising the unconfined yield strength at a range of applied loads. For applied stresses below 1 kPa, it becomes increasingly difficult to obtain reliable values of the unconfined yield strength. The bulk cohesion and tensile strength of the powder are then obtained by extrapolating the yield locus to zero and negative loads, respectively. However, the reliability of this approximation for a given material is not known. To overcome this limitation, techniques such as the Raining Bed Method, Sevilla Powder Tester and the newly-developed Ball Indentation Method may be used. In this paper, we report our measurement results of the tensile strength of glass beads, a-lactose monohydrate and various sizes of fluid catalytic cracking powders determined by the Sevilla Powder Tester and Raining Bed Method and compare them with those inferred from the Schulze Shear Cell. The results of the latter are also compared with those of the Ball Indentation Method. The outcome suggests that in the case of shear cell tests, the extrapolation of the yield locus to lower or negative loads is unsafe. The ball indentation method enables the characterisation of highly cohesive powders at very low compressive loads; however extrapolation to negative loads is still not reliable. In contrast, the Sevilla Powder Tester and Raining Bed Methods are able to characterise the tensile strength directly, but high bulk cohesion poses difficulties as the internal bed failure needs to be analysed in order to reliably estimate the tensile strength. These methods provide a better understanding of powder flow behaviour at low stresses, thus enabling a greater control of manufacturing processes. (C) 2015 Elsevier B.V. All rights reserved.